Thank you for the excellent work on Align3R — it’s truly impressive and very helpful for my research. I’m currently working on 4D reconstruction and using Align3R as one of the baselines.
While studying the implementation, I noticed that the depth extraction process is extremely slow. For example, extracting depth for the full Tartanair dataset seems to take over 90 days on a single RTX 3090, which seems unusually long.
Is this expected behavior?
Could you share how many GPUs you used and how long it took to preprocess the training data on your side?
Additionally, if you’re open to it, I’d really appreciate having a way to reach out (e.g., WeChat) in case I run into further technical issues.
Thanks again for your time and contribution!
Best regards,
Haosen Yang
Thank you for the excellent work on Align3R — it’s truly impressive and very helpful for my research. I’m currently working on 4D reconstruction and using Align3R as one of the baselines.
While studying the implementation, I noticed that the depth extraction process is extremely slow. For example, extracting depth for the full Tartanair dataset seems to take over 90 days on a single RTX 3090, which seems unusually long.
Is this expected behavior?
Could you share how many GPUs you used and how long it took to preprocess the training data on your side?
Additionally, if you’re open to it, I’d really appreciate having a way to reach out (e.g., WeChat) in case I run into further technical issues.
Thanks again for your time and contribution!
Best regards,
Haosen Yang